When a charged particle beam scanning microscope such as a scanning electron microscope (SEM) is used to measure the fine three-dimensional shape of the surface of a pattern of semiconductors, a charged particle beam must be obliquely routed to a specimen. The specimen is scanned in a plurality of directions with beams whose incidence angles are largely different from one another. Pieces of information on the height and slope of the surface of the specimen are acquired from secondary-electron images.
For the above measurement, a method of mechanically tilting a specimen and a method of holding a specimen horizontally and electromagnetically bending a charged particle beam are currently adopted. In equipment in which a voltage is applied to a specimen in order to induce a decelerating electric field in the space spreading in front of the specimen, and a beam is forcibly converged in order to attain a high resolution ranging from 3 nm to 5 nm, when the specimen is tilted, the decelerating electric field is so deformed that a normal high-resolution image cannot be formed. Since a field of view may be displaced due to the tilt of the specimen and it takes much time to mechanically tilt the specimen, the throughput of measurement is degraded. The method of tilting the charged particle beam is advantageous in terms of quickness and reproducibility. However, when a beam is tilted by an angle that is ten times larger than a normal convergence angle in order to observe a three-dimensional shape, the adverse effect of aberration caused by an objective lens is intensified. This results in a displacement of a field of view, degradation of a resolution derived from blurring of a spot, or a stream of an image. The resolution becomes much lower than the one attained when a beam is routed normally at right angles.
In efforts to minimize the blurring or displacement of a spot attributable to tilt of a beam, a method of introducing a deflection unit into a field induced by an objective lens in order to attain a desired tilt angle so as to cancel magnification chromatic aberration, that is, chromatic aberration causing enlargement has been proposed (refer to, for example, Japanese Patent Application Laid-Open No. 2000-348658).
Moreover, a method for compensating blurring of an image derived from the magnification chromatic aberration in which a Wiener filter installed in front of an objective lens is used to disperse energy according to an angle, by which a beam is tilted, to such an extent that the magnification chromatic aberration of the objective lens can be canceled has been proposed (refer to, for example, Japanese Patent Application Laid-Open No. 2001-15055).
Moreover, a method of canceling aberration, which an objective lens causes when a beam is tilted, by utilizing the off-axis aberration of other lens (refer to, for example, Japanese Patent Application Laid-Open No. 2004-127930) and a method of electromagnetically tilting an objective lens (refer to, for example, Japanese Patent Application Laid-Open No. 2003-168383 have been proposed.
Moreover, from the viewpoint of compensation of aberration, a means for convoluting a quadrupole field or an octupole field to a charged particle beam in the vicinity of an objective lens where the distance of the beam from an optical axis is large and compensation is achieved quite effectively so as to cancel the chromatic aberration or spherical aberration of the objective lens occurring in a direction corresponding to a certain direction of tilt in which the charged particle beam is tilted, and thus preventing degradation of a resolution has been proposed (refer to, for example, U.S. Pat. No. 6,627,890).